confusing locations of spatial points with periodic conditions

[YichengDWu]

The derived ODE problem below contains 5 spatial points, from the first inner point to the right boundary. I believe it's a natural result of structural simplification. But the users need to be notified the points are located from the first inner point to the right boundary, not the other way around

using ModelingToolkit, MethodOfLines, DomainSets
using OrdinaryDiffEq

@parameters t x
@variables u(..) 

Dt = Differential(t)
Dx = Differential(x)

eq = Dt(u(t, x)) ~  -2.0  * Dx(u(t, x))

tmin,tmax = 0, 4
xmin,xmax = 0, 1

domains = [x ∈ Interval(xmin, xmax),
           t ∈ Interval(tmin, tmax)]

bcs = [u(0,x) ~ sin(2π*x),
       u(t,xmin)~u(t,xmax)]
       
@named pdesys = PDESystem(eq,bcs,domains,[t,x],[u(t,x)])

discretization = MOLFiniteDifference([x=>0.2], t, approx_order=2, grid_align=center_align)
prob = discretize(pdesys,discretization)

@show prob.u0

grid = get_discrete(pdesys,discretization)
grid[u(t,x)]

[xtalax]

Please retrieve the solution like this:
solu = [map(d -> sol[d][i], grid[u(t, x)]) for i in 1:length(sol[t])]

There are no guarantees on the shape of the raw solution after structural_simplify

[YichengDWu]

Please retrieve the solution like this: solu = [map(d -> sol[d][i], grid[u(t, x)]) for i in 1:length(sol[t])]

There are no guarantees on the shape of the raw solution after structural_simplify

This is a bit cumbersome. I am satisfied with sol[grid[u(t,x)]]. Guess we just have to save the ODE prob and the grid too. Thanks!

[xtalax]

This is true until we have the solution interface, which will handle this automatically. That indexing may be better, I will see if I can update the docs to use that.